linux-kernel - RE: sata_sil24 memory fragmentation issues / VM allocation failures

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <5ABE8A5E96C0364CAF2B4DC1DEB7CEAC1282A527@Mercury.usurf.usu.edu>
Date:	Fri, 3 Sep 2010 20:10:44 +0000
From:	Jonathan Haws <Jonathan.Haws@....usu.edu>
To:	Robert Hancock <hancockrwd@...il.com>
CC:	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: sata_sil24 memory fragmentation issues / VM allocation failures

> I meant that "sata" is just the process name (I assume), it's not
> really anything to do with the SATA driver. Normally SATA host
> controller drivers don't really allocate memory at runtime so this
> wouldn't really be an issue with them. Network controllers do in order
> to handle received packets, though - it appears that for some reason
> the memory allocation by the network driver is failing.

You are correct - "sata" is the name of my SATA benchmarking app.  I had the same thought, so I went to the network driver and changed it so that it only ever dealt with single pages - the motivation being that if it only dealt with single pages I would never have a page allocation failure unless I was completely out of memory.  I made those changes and ended up seeing the same errors.

Someone did mention to me that the VM keeps an emergency reserve around for emergency allocations, so maybe I am hitting that barrier?

The thing that really gets me is that when I run the network alone or the disk alone there are no issues at all.  It is only when I run them together that I see problems.  I tried increasing how many packets I would handle in the polling function for the network, but that seemed to have zero effect.

> I'm not really sure why that is - it seems like you do have memory
> available. Hopefully some VM guru can pipe up with a suggestion.

That is what I am hoping as well.  Anyone have any thoughts?

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/